DKA Management: What Changed in 2024–2026? An Evidence-Based Update for Hospitalists

Updated September 2026

Diabetic ketoacidosis (DKA) is one of the most familiar emergencies in hospital medicine. The basic treatment—fluids, insulin, electrolyte management, and treatment of the precipitating cause—has not changed.

But several important details have.

From the 2024 international consensus report through randomized trials and meta-analyses published in 2025–2026, the evidence is refining how hospitalists diagnose DKA, choose IV fluids, administer basal and subcutaneous insulin, manage potassium, and recognize SGLT2 inhibitor-associated euglycemic DKA.

Here are the changes worth knowing in 2026.

1. DKA Diagnosis Has Become More Ketone-Focused

One of the most important conceptual changes is greater emphasis on direct measurement of β-hydroxybutyrate rather than relying primarily on the anion gap or urine ketones.

The 2024 international consensus report defines DKA using three components:

Diabetes/hyperglycemia: glucose ≥200 mg/dL (11.1 mmol/L) or a prior history of diabetes

Ketosis: β-hydroxybutyrate ≥3.0 mmol/L or urine ketones ≥2+

Metabolic acidosis: pH <7.3 and/or bicarbonate <18 mmol/L

The lower glucose threshold is particularly important because DKA does not always present with dramatic hyperglycemia.

That matters increasingly in the era of SGLT2 inhibitors.

Hospitalist pearl

Do not exclude DKA because the glucose is “only” 180 mg/dL.

If the clinical picture fits—particularly in a patient taking an SGLT2 inhibitor—check β-hydroxybutyrate and acid-base status.

2. Balanced Crystalloids vs Normal Saline: The Story Is More Nuanced

For decades, 0.9% saline was the default initial fluid for DKA.

The problem is chloride.

Large-volume normal saline administration can produce hyperchloremic metabolic acidosis, potentially making the laboratory picture look as though acidosis is persisting even after ketoacidosis is improving.

The 2024 consensus report recognizes either isotonic saline or balanced crystalloid solutions as reasonable initial resuscitation fluids.

In adults without cardiac or renal compromise, an initial rate around 500–1,000 mL/hour for the first 2–4 hours can be used, followed by individualized replacement based on hemodynamics, sodium, fluid balance, and comorbidities.

What did the newer RCT show?

A randomized trial involving 88 patients with severe DKA compared Ringer’s lactate with 0.9% saline.

Ringer’s lactate did not significantly shorten the time to DKA resolution.

However, patients receiving Ringer’s lactate had:

Lower chloride levels
Better potassium profiles
Less hyperchloremia
Less hypokalemia
Hospitalist pearl

Balanced crystalloids are a very reasonable choice, particularly when large-volume resuscitation is anticipated.

But the latest randomized evidence does not establish that lactated Ringer’s dramatically accelerates DKA resolution.

Think:

“Less chloride burden” rather than “magic DKA fluid.”

3. Early Basal Insulin May Be the Most Important Practical Update

Traditionally, many clinicians waited until DKA had resolved before administering long-acting insulin.

Increasing evidence supports another strategy:

Give basal insulin while the IV insulin infusion is still running.

The 2024 consensus report acknowledges evidence supporting basal insulin at approximately:

0.15–0.3 units/kg subcutaneously

during IV insulin treatment.

Why?

IV regular insulin has an extremely short duration of action. When the infusion stops, insulin activity can disappear quickly.

Long-acting insulin creates an insulin “bridge,” potentially reducing rebound hyperglycemia and facilitating transition off the infusion.

The 2025 Randomized Trial

A randomized clinical trial enrolled 90 adults with type 1 diabetes and DKA.

Patients received standard IV regular insulin alone or IV insulin plus early:

Glargine U100: 0.3 units/kg

or

Glargine U300: 0.3 units/kg

The results were striking.

Mean time to DKA resolution was approximately:

Glargine U100: 6.78 hours

Glargine U300: 6.33 hours

IV insulin alone: 11.12 hours

The difference was statistically significant.

Total insulin requirements were also lower in the early-basal groups.

Importantly, the investigators did not find significant differences among groups in:

Hypoglycemia
Hypokalemia
Rebound hyperglycemia
Recurrent DKA
4. The 2026 Meta-Analysis Strengthens the Basal-Insulin Argument

The evidence became even more interesting in 2026.

A systematic review and meta-analysis of eight randomized controlled trials involving 407 adults with DKA evaluated early subcutaneous basal insulin combined with IV insulin.

Early basal insulin reduced time to DKA resolution by approximately:

4.1 hours

compared with IV insulin alone.

It also modestly reduced fluid requirements.

Importantly, investigators found no significant increase in:

Hypoglycemia
Hypokalemia
Rebound hyperglycemia
Recurrent DKA
Mortality

Hospital length of stay was not significantly reduced.

What does this mean clinically?

Early basal insulin is no longer simply a pharmacologically attractive idea.

Multiple randomized trials now support the strategy.

The evidence is not perfect—many trials remain relatively small, and large multicenter outcome trials are still desirable—but this is increasingly reasonable evidence-based practice.

Hospitalist pearl

For an appropriate adult receiving an IV insulin infusion for DKA, consider giving the patient’s basal insulin early rather than waiting until the moment of transition.

If initiating basal insulin, current evidence and consensus guidance generally involve doses in the 0.15–0.3 units/kg range, individualized for insulin sensitivity, renal function, nutritional status, and hypoglycemia risk.

5. Not Every DKA Patient Needs an IV Insulin Infusion

Another important change involves subcutaneous rapid-acting insulin.

Traditional teaching was simple:

DKA = insulin drip.

That is no longer universally true.

For uncomplicated mild or moderate DKA, subcutaneous rapid-acting insulin analogs administered every 1–2 hours can be an effective alternative to IV insulin.

This strategy may allow appropriately selected patients to be managed in an emergency department or monitored medical/step-down environment without ICU admission solely for an insulin infusion.

Who should NOT be managed this way?

SC insulin protocols are not appropriate substitutes for IV insulin in patients with:

Severe DKA
Hemodynamic instability
Significant altered mental status
Critical illness
Significant DKA/HHS overlap
Situations where frequent monitoring cannot be provided
Hospitalist pearl

DKA does not automatically equal ICU.

Severity, clinical status, nursing capability, monitoring resources, and institutional protocols should determine the appropriate level of care.

6. Glucose Is Not the Endpoint of DKA Treatment

This remains one of the most important bedside teaching points.

Glucose frequently normalizes before ketoacidosis resolves.

When glucose falls below approximately 250 mg/dL, the answer is generally not to stop insulin.

Instead:

Add dextrose.

The 2024 consensus recommends adding approximately 5–10% dextrose and reducing the IV insulin infusion to approximately:

0.05 units/kg/hour

Insulin is then continued to suppress ketogenesis until DKA has resolved.

Hospitalist pearl

Treat the ketones and acidosis—not just the glucose.

Dextrose allows insulin to continue safely after serum glucose has normalized.

7. Potassium: A Small Threshold Change With Major Consequences

DKA patients are usually total-body potassium depleted, even when the initial serum potassium concentration is normal or elevated.

Insulin, correction of acidosis, volume expansion, and urinary losses can rapidly decrease serum potassium.

The 2024 consensus recommends beginning potassium replacement when serum potassium falls below:

5.0 mmol/L

with a target concentration around:

4–5 mmol/L.

For many patients, approximately 20–30 mmol of potassium per liter of IV fluid is sufficient, although replacement must be individualized.

The critical threshold is:

K⁺ <3.5 mmol/L

If potassium is below 3.5 mmol/L:

Replace potassium first and delay insulin until K⁺ rises above 3.5 mmol/L.

A suggested replacement rate in the consensus report is approximately 10 mmol/hour, with adjustment according to clinical circumstances and institutional protocols.

Monitoring

Check potassium at baseline, approximately 2 hours after starting insulin, and then approximately every 4 hours until DKA resolves.

Why does this matter?

Severe treatment-associated hypokalemia is not benign.

It is associated with substantially increased mortality.

Hospitalist pearl

Before starting insulin, ask:

“What is the potassium?”

Not simply:

“What is the glucose?”

8. Bicarbonate: Still Almost Never

Despite decades of debate, routine bicarbonate therapy remains unsupported.

Fluid resuscitation and insulin correct the underlying ketoacidosis.

Studies have not demonstrated meaningful improvement in cardiac or neurologic outcomes or faster recovery from DKA with routine bicarbonate.

Potential harms include:

Hypokalemia
Reduced tissue oxygen uptake
Cerebral edema
Paradoxical CNS acidosis
When should bicarbonate be considered?

The current consensus recommends considering bicarbonate for severe metabolic acidosis with pH <7.0.

If used, one suggested regimen is:

100 mmol sodium bicarbonate in 400 mL sterile water every 2 hours until pH >7.0.

Hospitalist pearl

Bicarbonate is rescue therapy for extreme acidemia—not routine DKA therapy.

9. SGLT2 Inhibitors Have Changed the DKA Differential

One of the biggest changes in modern DKA recognition is the widespread use of:

Empagliflozin
Dapagliflozin
Canagliflozin
Other SGLT2 inhibitors

These medications provide major cardiovascular and renal benefits in appropriate patients.

But they also introduce an important diagnostic problem:

Euglycemic DKA.

A patient can have significant ketoacidosis without severe hyperglycemia.

The 2024 consensus describes euglycemic DKA as DKA occurring with glucose concentrations <200 mg/dL.

Potential triggers include:

Acute illness
Surgery
Prolonged fasting
Reduced carbohydrate intake
Dehydration
Insulin reduction or omission
Excessive alcohol intake
The diagnostic trap

A patient presents with:

Glucose 168 mg/dL

Bicarbonate 12 mmol/L

Anion gap 23

Nausea and tachypnea

and is taking empagliflozin.

If the clinician thinks:

“The glucose isn’t high enough for DKA,”

the diagnosis may be missed.

Instead, think:

“Could this be euglycemic DKA?”

Check β-hydroxybutyrate.

10. Treat SGLT2-Associated Euglycemic DKA Like DKA—But Give Dextrose Earlier

The fundamental treatment remains:

Fluids + insulin + electrolyte management + treatment of the precipitating cause.

The SGLT2 inhibitor should be stopped during the acute episode.

Because glucose may already be normal or only mildly elevated, however, dextrose often needs to begin earlier.

The 2024 consensus recommends adding 5–10% dextrose to IV fluids from the beginning when treating euglycemic DKA so insulin can suppress ketogenesis without causing hypoglycemia.

Hospitalist pearl

In euglycemic DKA:

Dextrose does not treat the DKA.

It enables you to give the insulin that does.

11. Stop Using the Anion Gap Alone to Decide When DKA Has Resolved

Another subtle but important change involves determining resolution.

Historically, clinicians often followed closure of the anion gap.

But chloride-rich fluid administration can produce a persistent hyperchloremic non-anion-gap metabolic acidosis during recovery.

Conversely, other unmeasured anions can affect the gap.

The modern approach emphasizes β-hydroxybutyrate and acid-base status.

DKA resolution is generally defined by:

Plasma β-hydroxybutyrate <0.6 mmol/L

plus:

Venous pH ≥7.3 OR bicarbonate ≥18 mmol/L

Ideally, glucose should also be <200 mg/dL.

Urine ketones should not be used to determine resolution because β-hydroxybutyrate is converted back to acetoacetate during recovery, potentially making urine ketones remain positive—or even appear worse—despite clinical improvement.

Hospitalist pearl

Don’t chase the urine ketones.

And don’t let saline-induced hyperchloremic acidosis unnecessarily prolong an insulin infusion when the ketoacidosis itself has resolved.

12. Transitioning Off IV Insulin: Overlap Matters

Never simply stop an insulin infusion and then order basal insulin.

Long-acting insulin requires time to take effect.

The consensus recommendation is to administer basal insulin at least:

1–2 hours before discontinuing IV insulin.

An estimated total daily insulin requirement can be derived from:

Weight
Previous outpatient insulin requirements
Inpatient insulin requirements

A basal-bolus regimen is generally preferred.

Patients already receiving basal insulin before admission can often continue their basal insulin during DKA treatment, with appropriate adjustment.

The 2026 DKA Algorithm
ONE-SLIDE HOSPITALIST APPROACH
1 → CONFIRM DKA

Glucose ≥200 mg/dL OR known diabetes

PLUS

β-hydroxybutyrate ≥3 mmol/L

PLUS

pH <7.3 and/or HCO₃⁻ <18 mmol/L

↓

2 → START FLUID

Balanced crystalloid OR 0.9% saline

Typically 500–1,000 mL/hour initially in adults without cardiac/renal compromise.

Then individualize.

↓

3 → CHECK POTASSIUM BEFORE INSULIN

K⁺ <3.5

→ HOLD insulin
→ Replace K⁺
→ Start insulin when K⁺ >3.5

K⁺ 3.5–5.0

→ Start insulin
→ Add potassium to IV fluids as appropriate

K⁺ >5.0

→ Start insulin
→ Monitor closely

Target:

K⁺ 4–5 mmol/L

↓

4 → GIVE INSULIN

Moderate/severe DKA:

IV regular insulin ~0.1 U/kg/hour

Consider early basal insulin:

~0.15–0.3 U/kg SC

For selected uncomplicated mild/moderate DKA:

SC rapid-acting insulin every 1–2 hours may be appropriate

↓

5 → GLUCOSE <250?

YES → ADD 5–10% DEXTROSE

Reduce IV insulin approximately to:

0.05 U/kg/hour

Continue insulin until ketoacidosis, not merely hyperglycemia, resolves.

↓

6 → BICARBONATE?

Usually:

NO

Consider only with:

pH <7.0

↓

7 → CHECK FOR SGLT2 INHIBITOR

Normal-ish glucose does NOT exclude DKA.

Suspect euglycemic DKA.

Check β-hydroxybutyrate.

Stop SGLT2 inhibitor during acute DKA and begin dextrose early enough to permit continued insulin therapy.

↓

8 → DKA RESOLVED?

β-hydroxybutyrate <0.6 mmol/L

AND

pH ≥7.3 OR HCO₃⁻ ≥18 mmol/L

↓

9 → TRANSITION

Give/ensure basal insulin coverage.

Overlap basal insulin with IV insulin by at least 1–2 hours before stopping the infusion.

Five Changes Hospitalists Should Remember

If you remember only five things from the 2024–2026 DKA literature:

1. Measure β-hydroxybutyrate.

DKA diagnosis and resolution are increasingly ketone-centered rather than dependent on anion-gap calculations.

2. Balanced crystalloids are reasonable.

They may reduce hyperchloremia, although the newest RCT did not demonstrate faster DKA resolution with Ringer’s lactate.

3. Consider basal insulin early.

Randomized evidence increasingly supports giving basal insulin during the IV insulin infusion, with meta-analysis suggesting approximately four hours faster DKA resolution.

4. Mild/moderate uncomplicated DKA does not always require an insulin drip.

Selected patients can be treated with frequent subcutaneous rapid-acting insulin using an appropriate protocol and monitoring environment.

5. Think euglycemic DKA.

Especially when the patient is taking an SGLT2 inhibitor.

A glucose of 150–200 mg/dL does not rule out life-threatening ketoacidosis.

The Bottom Line

The fundamentals of DKA treatment have not been overturned.

The modern approach is becoming more precise.

Use β-hydroxybutyrate to diagnose and follow ketoacidosis.

Consider balanced crystalloids to reduce chloride burden.

Don’t stop insulin simply because glucose normalizes—add dextrose and keep treating ketogenesis.

Pay meticulous attention to potassium.

Avoid routine bicarbonate.

Consider early basal insulin while the IV infusion is running.

Recognize that selected patients with uncomplicated mild or moderate DKA may be candidates for subcutaneous rapid-acting insulin protocols.

And whenever a patient taking an SGLT2 inhibitor develops unexplained high-anion-gap metabolic acidosis, nausea, vomiting, or tachypnea:

Think DKA—even when the glucose doesn’t look like DKA.
References & Further Reading

Umpierrez GE, et al. Hyperglycemic Crises in Adults With Diabetes: A Consensus Report. Diabetes Care. 2024.

American Diabetes Association Professional Practice Committee. Diabetes Care in the Hospital: Standards of Care in Diabetes—2026. Diabetes Care. 2026.

El Feky AY, et al. Effectiveness and Safety of Early Insulin Glargine U100 and Glargine U300 Administration in the Management of Diabetic Ketoacidosis in Adults With Type 1 Diabetes Mellitus: A Randomized Clinical Trial. Endocrine Practice. 2025.

AbuJlambo A, et al. Impact of Early Subcutaneous Basal Insulin With Intravenous Insulin Infusion for Diabetic Ketoacidosis and Glycaemic Outcomes: A Systematic Review and Meta-Analysis. Endocrinology, Diabetes & Metabolism. 2026.

Trifi A, et al. 0.9% Sodium Chloride Versus Ringer’s Lactate in the Management of Severe Diabetic Ketoacidosis: A Randomized Trial. Medicina Intensiva. 2026